Never re-decide something your team already settled.

Catches the conflict before it ships, not after.

AI Oracle reads your team's decisions, notes, and prior threads, and surfaces the right one the moment someone's about to contradict it, right in the doc, sheet, or slide they're writing.

Free during early access. Sign in with email.

AI Oracle
Drafting the renewal quote
“...to close this out, we can offer a 30% discount on the renewal...”
checked instantly
Flagged before it sent
Conflict · discount cap decision

Your team ratified a 20% discount cap in March. This quote is 30%. Here's the decision and who approved it.

Which of these is you

Most teams that try this recognize themselves in more than one.

Small team founder

“I'm the only one who remembers why we decided that.”

Around hire four or five, someone re-litigates a call the team already made six months ago.

Ops, RevOps, or product marketing

“Different teams, different quarters, conflicting commitments.”

The platform team picks one direction in Q2, the mobile team picks the opposite in Q4, and nobody finds out until a customer notices.

Deal desk, GC, or compliance

“I need the record to exist before the audit, not after.”

Catch the off-policy quote or claim at write-time, with a defensible trail of what was flagged and what happened next.

Before anything else

Your data is yours. Full stop.

Mark something personal and it's yours alone — not even org admins can read it. Mark it business and your team can draw on it while you're there. You decide which is which, every time, and you can change it later.

Delete your account and everything goes with it — every note, every decision, every connected token, removed from our database within 30 days. Not anonymized and kept. Gone.

The decision was made. The note exists. The explainer was written. Six months later, in a different doc, a different person solves the same problem from scratch, or contradicts it outright.

None of it is gone. Nobody remembered to look. Wikis assume people browse them. Search assumes people remember something exists and go find it. Neither happens in the middle of writing a proposal.

What you get back

Every check tells you what kind of match it found.

Not ten search results to sort through. One typed signal, so your team acts on what matters, not whatever was nearby.

Search gives you ten links. AI Oracle tells you which one matters and why.

Conflict
Contradicts a decision your team already ratified
VIOLATES · TEAM DECISION

“Discount cap is 20%. You've quoted 30%.”

Support
Backed by work your team already did
SUPPORTS · TEAM DECISION

“This matches the Q3 pricing decision — here's who ratified it.”

Education
Someone already wrote the explainer
RELATED · TEAM NOTE

“This came up on the March customer call. Here's what was said.”

History
The reasoning behind the policy, not just the rule
RELATED · DECISION LOG

“Here's why the 20% cap was set, and who signed off on it.”

How it works

Save it once. Ask it forever.

The three things happening under the hood.

1 · Save it anyhow

Every door leads to one place.

TypeClipPasteUploadTalkRecord

Type it, clip it from any page, paste it, upload a doc, talk, or record. Every door funnels into one searchable archive.

2 · Ask it back

Plain English, not a search bar.

meaning + wordsblendde-dupethe one answer

It runs a two-pass search (meaning and keywords), blends them, bounces duplicates, and picks the one answer that matters.

3 · It gets sharper

A thumbs up or down, and it learns how you decide.

👍 / 👎calibrated to you

A thumbs up or down calibrates it to how you actually decide, with a guardrail so a correction never makes it worse.

Why trust this now

Ahead of a trend, not an unproven newcomer.

Every major AI assistant got a memory in 2026. None of them work outside their own app.

Zero prompts stored

What you check is evaluated and discarded, never persisted server-side beyond the request.

Zero-retention LLM contracts

The LLMs we call have published zero-retention policies on our usage. Not training data, not ours and not anyone else's.

Nothing promoted without you

AI Oracle proposes and surfaces. It never ratifies or applies anything on its own; every change is your explicit accept.

“I built this while recovering from a spinal injury, when typing wasn’t always an option. I needed something that worked with whatever I could manage that day: a voice memo, a video, a quick note, however I could get it down. Save it once, however you can, and it comes back when you need it. That’s still the whole idea.”
Adam, founder

Pricing

No seat minimum. A team of one counts.

A shared corpus your whole team writes against. Per seat, no minimum — a team of one counts.

Most popular

Team

$25/seat/mo

For people who share one corpus.

  • Unlimited decisions + notes
  • All four surfacing modes
  • Org-shared corpus & guardrails
  • Team memory tier
  • Team activity log, exportable as CSV
  • Slack bot (when shipped)
Start free → upgrade

Business

$50/seat/mo

For teams with a compliance hat to wear.

  • Everything in Team
  • Audit log — every check across the org, not just your own
  • Export findings as CSV or JSON
  • Priority support
Start free → upgrade

Enterprise

Custom

Regulated, on-prem, or 100+ seats — let's build it together.

  • Everything in Business
  • SOC 2 + BAA
  • SAML SSO & SCIM
  • On-prem / private LLM
  • Custom retention + scheduled exports
  • Dedicated support
Talk to our team

No credit card to start. Cancel anytime.

Asked and answered

The questions you're about to ask.

Isn't this just Confluence, Notion, or a team wiki?
Wikis are where you put things. Nobody flags it when a new page contradicts an old one, so the reader has to already know to go check. AI Oracle reads what you're actually writing and surfaces the conflict, or the supporting decision, without anyone remembering to look.
Do we need to migrate our docs before this is useful?
No. Bring what you have, or start from zero and let it build as you work. The extension captures as your team browses, notes get created in-app, a Notion or Obsidian export imports directly, and prior decisions can be pasted in. One ratified decision is enough to start catching conflicts against it.
What happens to our data if we cancel, or delete the account?
If you stop paying, nothing is deleted: the org drops to Free-tier limits on new items, but everything already saved stays exactly as it was. If you delete the account, all content and connected tokens are deleted within 30 days. Org-wide activity exports as CSV any time. Full detail at /privacy.
Can people keep personal notes separate from the team's knowledge?
Yes, per item, chosen by whoever saves it. Mark something personal and it's visible only to them, not even org admins can read it. Mark it business and the team can draw on it. Either can change later.
Do we need a minimum number of seats?
No seat minimum on Team or Business. A team of one counts, and seats add or drop as your org changes.
Is this only useful once a decision has already gone wrong?
No — that's the point of catching it at write-time instead of at review or audit. AI Oracle checks the draft before it ships, not the record after the fact, so the contradiction gets caught in the moment someone's about to make it, not months later when someone stumbles on it.

Start free. No credit card. Bring what you have. We'll start from there.

Start free, no credit card